checkpoint model on first step callback (#2906)

* checkpoint model on first step callback

* remove debug

* add test cases; update existing tests not to save on first step

* move test out of solo

* delete

* default to False

* typo

examples/phi/lora-3.5.yaml

@@ -63,3 +63,5 @@ warmup_steps: 10
 evals_per_epoch: 4
 saves_per_epoch: 4
 weight_decay: 0.0
+
+# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/phi/phi-ft.yml

@@ -57,3 +57,5 @@ weight_decay: 0.1
 resize_token_embeddings_to_32x: true
 special_tokens:
   pad_token: "<|endoftext|>"
+
+# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/phi/phi-qlora.yml

@@ -60,3 +60,5 @@ weight_decay: 0.1
 resize_token_embeddings_to_32x: true
 special_tokens:
   pad_token: "<|endoftext|>"
+
+# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/phi/phi2-ft.yml

@@ -57,3 +57,5 @@ weight_decay: 0.1
 resize_token_embeddings_to_32x: true
 special_tokens:
   pad_token: "<|endoftext|>"
+
+# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/phi/phi3-ft-fsdp.yml

@@ -71,3 +71,5 @@ fsdp_config:
 resize_token_embeddings_to_32x: true
 special_tokens:
   pad_token: "<|endoftext|>"
+
+# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/phi/phi3-ft.yml

@@ -59,3 +59,5 @@ warmup_ratio: 0.2
 debug: true
 weight_decay: 0.1
 resize_token_embeddings_to_32x: true
+
+# save_first_step: true  # uncomment this to validate checkpoint saving works with your config